Durante las últimas décadas se ha realizado mucho esfuerzo para encontrar soluciones eficientes al problema de detección de los radares marítimos modernos. La probabilidad de detección de objetos de interés depende altamente de los esquemas de detección seleccionados y sus distribuciones de clutter y señal. Se han desarrollado diferentes estrategias de detección con el objetivo de discriminar el retorno marino de la señal de interés. Estas estrategias establecen un umbral que depende de la potencia del retorno marino local basada en un test de hipótesis para una determinada falsa alarma. En esta tesis, se presentan algunos detectores radar con el propósito de encontrar soluciones eficientes al problema de discriminar entre los objetos de interés y el ambiente marino, especialmente para el caso de condiciones no homogéneas que usualmente acaecen en el escenario marino. Con ese propósito, se describen algunos modelos del entorno marino por medio de sus distribuciones estadísticas para baja y alta resolución. Por ejemplo, para el caso de baja resolución y ángulos de incidencia altos, el clutter generalmente se modela por medio de la distribución Gaussiana. Sin embargo, para el caso de radares modernos de alta resolución, donde las celdas de rango son pequeñas con muy pocos dispersores, el retorno radar se aleja de las distribuciones Gaussianas y el ambiente marino se representa muy bien mediante distribuciones de cola larga, tales como las distribuciones Weibull y K. En este trabajo, se discuten los principios básicos de la detección radar. Se realiza una descripción detallada de algunos esquemas clásicos, tales como el detector de Neyman Pearson, el GLRT, y algunos esquemas de detección, como el CA CFAR, el GO CFAR, el SO CFAR, el OS CFAR y el detector propuesto por [60] . Teniendo en cuenta que la detección radar se puede interpretar como un problema de reconocimiento de patrones, las redes neuronales son apropiadas para tratarlo, por consiguiente, se consideraron algunas soluciones por medio de varios algoritmos tales como el ANN CFAR, el CANN CFAR y el NNCAOS CFAR. Finalmente, se presenta un nuevo detector radar en el contexto de los sistemas no-coherentes que operan en ambientes marinos no homogéneos. Se considera el caso especial de un radar de alta resolución con un ángulo de incidencia bajo, que funciona en condiciones de mar adversas. El esquema de detección que se plantea, se basa en la propuesta de un nuevo modelo del ambiente marino. Los objetivos de diseño de esta distribución fueron introducir grados de libertad adicionales (parámetros) con respecto a la distribución K para modelar los casos de colas pesadas y también obtener un detector en forma cerrada que aproxime la propiedad CFAR cuando trabaja en condiciones extremas. / During the last decades much effort has been done to find efficient solutions to the problem of detection for modern maritime radar. The probability of detection of a target highly depends on the selected detection scheme and its signal and clutter distributions. Different detection strategies have been developed in order to discriminate the clutter from the signal of interest. These detection strategies set a threshold that depends on the local clutter power based on a hypothesis test for an expected false alarm rate. In this thesis, several radar detectors were introduced with the purpose of finding efficient solutions to the problem of discriminating between targets and the clutter environment, especially for the case of non-homogeneous situations that usually happen in the sea scenario. To that purpose some clutter models are described by means of their statistical distributions for low and high resolution. For example, for the case of low resolution and high grazing angles the clutter is generally modeled by means of the Gaussian distribution. However, for the case of the modern high resolution radars, where the range cells are small with only few scatterers, the clutter models depart from the Gaussian distributions and the sea environments are well represented by means of long tailed distributions, like the Weibull and the K. In this work, the basic radar detection principles are discussed. A detailed description of some classic detection schemes are given, like the Neyman Pearson Detector, the GLRT, and some CFAR detection schemes, like the CA CFAR, the GO CFAR, the SO CFAR, the OS CFAR and the detector proposed by [60] . Considering that the radar detection may be interpreted as a pattern recognition, the Neural Networks are suitable to deal with the problem. Some solutions were considered by means of several algorithms like the specialized ANN CFAR, the CANN CFAR and the NNCAOS CFAR. Finally, a new radar detector in the context of non-coherent radar systems operating under non-homogeneous sea clutter and the special case of a high resolution radar at low grazing angles under adverse sea conditions, is presented. The novel detection scheme is based on a new clutter distribution model. The design objectives of this clutter distribution were to introduce additional degrees of freedom (parameters) with respect to the K distribution to model heavy tailed cases and also to obtain a closed-form detector that approximates the CFAR property when working under very stringent detection conditions.

Advances in digital signal processing (DSP) and computing technology have resulted in the emergence of increasingly adaptive radar systems. It is clear that the Electronic Attack (EA), or jamming, of such radar systems is expected to become a more difficult task. The reason for this research was to address the issue of jamming adaptive radar systems. This required consideration of adaptive jamming systems and the development of a methodology for outlining the features of such a system is proposed as the key contribution of this thesis. For the first time, game-based optimization methods have been applied to a maritime counter-surveillance/counter-targeting scenario involving conventional, as well as so-called ‘smart’ noise jamming.Conventional noise jamming methods feature prominently in the origins of radar electronic warfare, and are still widely implemented. They have been well studied, and are important for comparisons with coherent jamming techniques.Moreover, noise jamming is more readily applied with limited information support and is therefore germane to the problem of jamming adaptive radars; during theearly stages when the jammer tries to learn about the radar’s parameters and its own optimal actions.A radar and a jammer were considered as informed opponents ‘playing’ in a non-cooperative two-player, zero-sum game. The effects of jamming on the target detection performance of a radar using Constant False Alarm Rate (CFAR)processing were analyzed using a game theoretic approach for three cases: (1) Ungated Range Noise (URN), (2) Range-Gated Noise (RGN) and (3) False-Target (FT) jamming.Assuming a Swerling type II target in the presence of Rayleigh-distributed clutter, utility functions were described for Cell-Averaging (CA) and Order Statistic (OS) CFAR processors and the three cases of jamming. The analyses included optimizations of these utility functions, subject to certain constraints, with respectto control variables (strategies) in the jammer, such as jammer power and spatial extent of jamming, and control variables in the radar, such as threshold parameter and reference window size. The utility functions were evaluated over the players’ strategy sets and the resulting matrix-form games were solved for the optimal or ‘best response’ strategies of both the jammer and the radar.

Advances in digital signal processing (DSP) and computing technology have resulted in the emergence of increasingly adaptive radar systems. It is clear that the Electronic Attack (EA), or jamming, of such radar systems is expected to become a more difficult task. The reason for this research was to address the issue of jamming adaptive radar systems. This required consideration of adaptive jamming systems and the development of a methodology for outlining the features of such a system is proposed as the key contribution of this thesis. For the first time, game-based optimization methods have been applied to a maritime counter-surveillance/counter-targeting scenario involving conventional, as well as so-called ‘smart’ noise jamming.Conventional noise jamming methods feature prominently in the origins of radar electronic warfare, and are still widely implemented. They have been well studied, and are important for comparisons with coherent jamming techniques.Moreover, noise jamming is more readily applied with limited information support and is therefore germane to the problem of jamming adaptive radars; during theearly stages when the jammer tries to learn about the radar’s parameters and its own optimal actions.A radar and a jammer were considered as informed opponents ‘playing’ in a non-cooperative two-player, zero-sum game. The effects of jamming on the target detection performance of a radar using Constant False Alarm Rate (CFAR)processing were analyzed using a game theoretic approach for three cases: (1) Ungated Range Noise (URN), (2) Range-Gated Noise (RGN) and (3) False-Target (FT) jamming.Assuming a Swerling type II target in the presence of Rayleigh-distributed clutter, utility functions were described for Cell-Averaging (CA) and Order Statistic (OS) CFAR processors and the three cases of jamming. The analyses included optimizations of these utility functions, subject to certain constraints, with respectto control variables (strategies) in the jammer, such as jammer power and spatial extent of jamming, and control variables in the radar, such as threshold parameter and reference window size. The utility functions were evaluated over the players’ strategy sets and the resulting matrix-form games were solved for the optimal or ‘best response’ strategies of both the jammer and the radar.

In this thesis, improvement of Cell Averaging Constant False Alarm Rate (CA CFAR) radar processors is studied. A new improvement method is proposed that will reduce probability of false alarm while keeping probability of detection at good values. This method makes use of Doppler spreading patterns that appear after Doppler processing. Therefore this method is called Doppler Pattern Matching (DPM). Performance of the algorithm has been investigated by Monte Carlo simulations. In order to evaluate the performance, improvement factor is calculated which is the ratio of the probability of false alarm of original detector to the false alarm of improved detector. It is observed that improvement factor changes depending on the simulation scenario. Almost in every case, good improvement factor can be obtained. Moreover, in most of the cases, there has been no reduction in probability of detection after DPM is applied to CA CFAR detector.

Conventional methods of CFAR detection always use windowing, in the sense that some number of cells are investigated and the target present/absent decision is made according to the composition of the cells in that window. The most commonly used versions of CFAR detection algorithms are cell averaging CFAR, smallest of cell averaging CFAR, greatest of cell averaging CFAR and order-statistics CFAR. These methods all use windowing to set the decision threshold. In this thesis, rather than using windowed CFAR algorithms, a new method of estimating the background threshold is presented, analyzed and simulated. This new method is called the Switching IIR CFAR algorithm and uses two IIR filters to accurately estimate the background threshold. Then, using a comparison procedure, one of the filters is selected as the current threshold estimate and used. The results are seen to be satisfactory and comparable to conventional CFAR methods. The basic advantages of using the SIIR CFAR method are computational simplicity, small memory requirement and acceptable performance under clutter edges and multiple targets.

Constant False Alarm Rate (CFAR) Processors are important for applications where thousands of detection tests are made per second, such as in radar. This thesis introduces a new method for CFAR threshold estimation that is particularly applicable to sound source detection with distributed microphone systems. The novel CFAR Processor exploits the near symmetry about 0 for the acoustic pixel values created by steered-response coherent power in conjunction with a partial whitening preprocessor to estimate thresholds for positive values, which represent potential targets. To remove the low frequency components responsible for degrading CFAR performance, fixed and adaptive high-pass filters are applied. A relation is proposed and it tested the minimum high-pass cut-off frequency and the microphone geometry. Experimental results for linear, perimeter and planar arrays illustrate that for desired false alarm (FA) probabilities ranging from 10-1 and 10-6, a good CFAR performance can be achieved by modeling the coherent power with Chi-square and Weibull distributions and the ratio of desired over experimental FA probabilities can be limited within an order of magnitude.

The small target detection in High Frequency Surface Wave Radar is limited by the presence of various clutter and interference. Several novel signal processing techniques are developed to improve the system detection performance. As an external interference due to local lightning, impulsive noise increases the broadband noise level and then precludes the targets from detection. A new excision approach is proposed with modified linear predictions as the reconstruction solution. The system performance is further improved by de-noising the estimated covariance matrix through signal property mapping method. The existence of non-stationary sea clutter and ionospheric clutter can result in excessive false alarm rate through the high sidelobe level in adaptive beamforming. The optimum threshold discrete quadratic inequality constraints method is proposed to guarantee the sidelobe-controlling problem consistently feasible and optimal. This constrained optimization problem can be formulated into a second order cone problem with efficient mathematical solution. Both simulation and experimental results validate the improved performance and feasibility of our method. Based on the special noise characteristics of High Frequency radar, an adaptive switching Constant False Alarm Rate detector is proposed for targets detection in the beamformed range-Doppler map. The switching rule and adaptive footprint are applied to provide the optimum background noise estimation. For this new method about 14% probability of detection improvement has been verified by experimental data, and meanwhile the false alarm rate is reduced significantly compared to the original CFAR. The conventional Doppler processing has difficulty to recognize a target if its frequency is close to a Bragg line. One detector is proposed to solve this co-located co-channel resolvability problem under the assumption that target/clutter have different phase modulation. Moreover with the pre-whitening processing, the Reversible Jump Markov Chain Monte Carlo method can provide target number and Direction-of-Arrival estimation with lower detection threshold compared to beamforming and subspace methods. RJMCMC is able to convergent to the optimal resolution for a data set that is small compared with information theoretic criteria.

Les transmissions numériques par satellite sont largement utilisées dans plusieurs domaines allant des applications commerciales en orbites terrestres aux missions d'exploration scientifiques en espace lointain (Deep Space). Ces systèmes de transmission fonctionnent sur des très grandes distances et ils disposent des ressources énergétiques très limitées. Cela se traduit par un très faible rapport signal à bruit au niveau de la station de réception terrestre. Une possibilité d'établir une liaison fiable dans ces conditions très défavorables, réside dans l'utilisation de codes correcteurs d'erreurs puissants tels que les Turbo codes et le LDPC. Cependant, les gains de codage sont conditionnés par le bon fonctionnement des étages de la démodulation cohérente en amont, notamment l'étage de synchronisation. L'opération de synchronisation consiste à estimer et compenser le décalage en phase et en fréquence entre le signal reçu et l'oscillateur local du récepteur. Ces décalages sont généralement provoqués par des imperfections matérielles et le phénomène d'effet Doppler. A très faible rapport signal à bruit, les systèmes de synchronisation actuels se trouvent limités et incapables d'assurer les performances requises. Notre objectif est de fiabiliser l'étage de synchronisation du récepteur dans des conditions très difficiles de faible rapport signal sur bruit, d'effet Doppler conséquent avec prise en compte d'un phénomène d'accélération (Doppler rate) et d'une transmission sans pilote. Cette thèse CIFRE traite du problème de la synchronisation porteuse pour la voie descendante d'une transmission Deep Space. Après la réalisation d'une étude de l'état de l'art des techniques de synchronisation, nous avons retenu les boucles à verrouillage de phase (PLL: Phase Locked Loop). Dans un contexte industriel, les PLL offrent le meilleur compromis entre complexité d'implémentation et performances. Plusieurs détecteurs de phase basés le critère du maximum de vraisemblance ont été considérés et modélisés par leurs courbes caractéristiques. En se basant sur les modèles équivalents, nous avons développé une nouvelle étude de la phase d'acquisition non-linéaire d'une PLL du deuxième ordre avec un détecteur de phase semi-sinusoïdal. La deuxième partie de la thèse a été consacrée à l'étude des techniques de combinaison d'antennes. Ces méthodes visent à exploiter la diversité spatiale et améliorer le bilan de liaison de la chaîne de transmission tout en offrant une flexibilité de conception ainsi qu'une réduction considérable du coût d'installation. A l'issue de cette partie, nous avons proposé un nouveau schéma de combinaison d'antenne qui améliore le seuil de fonctionnement des systèmes existants. / In deep space communication systems, the long distance between the spacecraft and the ground station along with the limited capacity of the on-board power generator result a very low signal to noise ratio (SNR). However, such transmission still possible by using near Shannon limit error correction codes (Turbo code and LDPC code). Nevertheless, to take advantage of this coding gain, the coherent demodulation is mandatory, and the carrier phase synchronization must be reliable at more restrictive SNR. At very low SNR, current synchronization systems are limited and unable to provide the required performances. Our goal is to improve the reliability of the receiver synchronization stage under very difficult conditions of a very low SNR, a variable Doppler effect (Doppler rate) and a blind transmission. This thesis deals with the problem of carrier phase synchronization for the downlink of a Deep Space transmission. After the study of the existing solutions, we selected the phase locked loop (Phase Locked Loop: PLL). In an industrial context, PLL offers the best trade-off between complexity and performance. Several phase detectors based on the maximum likelihood criterion were considered and characterized by their S-curves. Based on the equivalent models, we have developed a new study of the non-linear acquisition phase of a second-order PLL with a semi-sinusoidal phase error detector. The second part of the thesis was dedicated to the antennas combining techniques. These methods aim to improve the link budget of the transmission and offer more flexibility. At the end of this part, we proposed a new antennas combining scheme that improves the operating threshold of existing systems.

In industrial environments workers can be prone to dangerous situations that may cause harm. In order to prevent accidents and fulfill legislative requirements for safe working condition, a safety system can be needed to be installed in place. In an attempt to bring a principal improvement in the safety standards offered by conventional systems in existence, a radar sensor is proposed for the detection of dangerous situations. This thesis explores the limiting performance of the radar sensor in an industrial safety system. The radar sensor is an FMCW mode MIMO radar and we develop a model to simulate the radar sensor based on its hardware design. The model is utilized to deduce and analyze the performance limits of the radar sensor, by using a Finite Element Method to determine the electromagnetic scattering response of a target in the operating frequency band, 60-64 GHz. Appropriate dielectric and magnetic properties of human tissue is assigned to obtain accurate scattering responses of a human body target. The performance analysis is done by comparing the reflection from various protruding body parts, with the reflection from the torso, over a wide range of postures of the body. Further, CFAR, a common form of detection algorithm is applied on a few of the postures to detect a protruding part the body. Finally, a discussion on the limited performance is made, and potential methods to enhance the performance of the radar sensor in an industrial safety system are suggested. / I industriella miljöer kan arbetare utsättas för farliga situationer vilka kan orsaka skador. För att förhindra olyckor och uppfylla lagstadgade krav för säkra arbetsförhållanden kan säkerhetssystem behöva installeras på plats. I ett försök att åstadkomma en förbättring av de säkerhetsstandarder som erbjuds av konventionella system föreslås en radarsensor för att detektera farliga situationer. Denna avhandling undersöker möjliga begränsningar i prestanda hos en radarsensor, vilken ingår som komponent i ett industriellt säkerhetssystem. Radarsensorn är en MIMO-radar, som arbetar i FMCW-mod, och vi utvecklar en modell för att simulera radarsensorn utgående ifrån dess hårdvarudesign. Modellen används för att härleda och analysera begränsningar i prestanda hos radarsensorn, genom att med hjälp av finita elementmetoden beräkna spridningen av elektromagnetiska vågor från ett radarmål i frekvensbandet 60-64 GHz. Tillämpliga dielektriska och magnetiska egenskaper hos mänsklig vävnad har använts för att erhålla tillförlitliga data på spridningsresponsen för människokroppen. Prestandaanalysen görs genom att jämföra reflektionen från olika utskjutande kroppsdelar med reflektionen från bålen över ett stort antal kroppshållningar. Vidare, CFAR, en vanlig form av detektionsalgoritm tillämpas på några av hållningarna för att detektera utskjutande delar av kroppen. Avslutningsvis diskuteras begränsningarna i prestandan och potentiella metoder för att förbättra radarsensorns prestanda i ett industriellt säkerhetssystem föreslås.

This thesis addresses the design study, implementation and analysis of signal processing algorithms for a 79 GHz millimeter-wave Phase Modulated Continuous Wave (PMCW) Multi Input Multi Output (MIMO) short range radar; performed in IMEC research institute (Leuven, Belgium). The radar system targets high resolution performance with low power consumption in order to integrate a full MIMO radar transceiver with digital processor and antennas in a compact package featuring a size of 1 cm2. Achieving such radar system characteristics requires the exploitation of a highly demanding digital architecture with signal processing gain and high range, speed and angle resolutions. The improved resolution and detection capabilities will be achieved by performing signal processing algorithms on the reflected waveform. The digital front-end implements parallel range gate processing with a bank of correlators that perform: pulse compression, coherent accumulation to further increase Signal to Noise Ratio (SNR) and N-point FFT to extract the Doppler information. The use of MIMO is proposed implementing a code domain technique in the PMCW waveform, the Outer Hadamard Code MIMO. This concept makes use of a unique sequence for all the transmitting antennas that is rendered by an outer sequence to ensure the orthogonality of the transmitted waveforms. The outer code makes use of the good cross-correlation properties of the Hadamard sequences and the waveform uses sequences that exhibit perfect auto-correlation profile, the Almost Perfect Autocorrelation Sequences (APAS). The MIMO implementation results in higher angular resolution and extra processing gain. The use of beamforming techniques in the radar allows the angle estimation of the detected targets; using rough and fine beamforming that provides with coarse and precise Angle of Arrival (AoA) estimation in an early and late stage respectively. A Constant False Alarm Rate (CFAR) processing stage is implemented in the stage of the system where higher signal processing gain is achieved. This algorithm allows the variation of the CFAR parameters and analyzes the detections in order to improve the probability of detection (Pd) while decreasing the probability of false alarm (Pfa). A series of simulations with different scenarios and variable parameters are set in order to analyze the performance of the system. The simulations analyze the gain achieved in each stage and their outcomes show an impressive processing gain that can reach SNR improvements as high as 77 dB for a small virtual array while keeping the Pfa low with the CFAR adjustment. The use of bigger arrays demonstrates the possibility to enable clear detections for low Radar Cross Section (RCS) targets in far distances of the unambiguous range. The use of beamforming shows interference reduction improvement as the beam widths narrow with the increasing number of virtual array antennas. These results have been achieved while keeping the system design parameters to a range resolution of 7.5 cm for a maximum range of 37.5 meters with speed resolution of 0.2 m/s and a maximum detectable speed of 12.66 m/s. The outcomes support the good performance of the signal processing techniques implemented and the benefits in applying them in a SoC mmWave MIMO radar.